长期演进系统在分组交换域内承载业务,延迟和分组丢弃敏感的实时业务服务质量(quality of service,QoS)难以得到满足。根据业务优先级差异、业务对延迟的敏感度等约束条件,提出了一种基于虚拟队列的分组延迟预测方法,进而根据分组延迟的预测结果确定其对系统资源需求的紧急程度,随后根据分组紧急度对缓存中的分组采用分类调度策略,以最大化系统资源利用率。结果表明,本文提出的调度策略能够有效地提高实时业务的QoS,改善网络的性能。
The long-term evolution system carries traffic in the packet-switched domain. The quality of service (QoS) of delay and packet loss-sensitive real-time services is difficult to be satisfied. According to the service priorities difference and the constraint condition of delay sensitivity, a virtual-queue-based packet delay prediction strategy is proposed. In the strategy, the packet urgency is determined by the predictable packet delay, and then the packets are classified according to the packet urgency. Compared with other scheduling strategies, results show that the scheduling strategy can improve the QoS of the real-time business effectively, and the network performance can be improved effectively.